HCASE

HCASE generates interpretable chemical-space embeddings of molecular structures by combining pseudo-Hilbert Curves with Scaffold-Keys derived from Bemis-Murcko scaffolds to support analysis of molecular relationships in medicinal chemistry.


Key Features:

  • Pseudo-Hilbert Curve Utilization: Uses pseudo-Hilbert Curves to embed structures while preserving spatial locality so similar molecules remain proximate in the embedded space.
  • Scaffold-Key Integration: Derives Scaffold-Keys from Bemis-Murcko scaffolds to anchor molecular structures within the defined chemical space.
  • Interpretable Embeddings: Produces embeddings that provide meaningful representations of molecular relationships and properties for medicinal chemistry analysis.

Scientific Applications:

  • Approved-drug analysis (DrugBank): Applied to analyze approved drug molecules from DrugBank within chemical spaces defined by Bemis-Murcko scaffolds extracted from ChEMBL v24.1.
  • Natural-product analysis (CANVASS): Applied to analyze natural products from CANVASS within chemical spaces defined by Bemis-Murcko scaffolds extracted from ChEMBL v23.

Methodology:

Extraction of Bemis-Murcko scaffolds, derivation and integration of Scaffold-Keys, and embedding those Scaffold-Keys into chemical space using a pseudo-Hilbert curve approach to produce spatially localized representations.

Topics

Details

License:
MIT
Tool Type:
command-line tool
Programming Languages:
Python, Shell
Added:
1/18/2021
Last Updated:
1/30/2021

Operations

Publications

Zahoranszky-Kohalmi G, Wan KK, Godfrey AG. Hilbert-Curve Assisted Structure Embedding Method. Unknown Journal. 2020. doi:10.26434/chemrxiv.11911296.v1.